MONTE analysis

This repository contains the code and notebooks used to reproduce the analyses presented in the MONTE paper. If you are looking for tutorial or the main package, please check MONTE.

---

Contents

• MONTE analysis
  • Contents
  • Folder structures
  • Environment setup
    • Python environment
    • Verify the installation
    • R environment
  • Running the analyses
  • Citation
  • Contact

Folder structures

Note

All data used in this study, including the TCGA and PCAWG datasets and generated files, are available on Zenodo. After downloading the archive, unzip it, and move data into this folder. The scripts should then be able to access all required files.

This repository contains two main folders: data and src. The scripts expect the required datasets to be available in the data folder. The src folder is divided into jupyter-notebooks, which contains the primary analysis code, and plots, which contains the scripts used to generate the figures. The notebooks corresponding to the analyses in the paper are listed in the table below. Because the plotting scripts are labeled with their corresponding figure numbers, they are not described further here.

Filename	Description	Corresponding section / figures in paper
01_monte_pancancer_model_training	Train the pan-cancer model and evaluate its performance on the test set	Pan-cancer model training
02_monte_model_ablation_study	Evaluate the contribution of each MONTE component	Figure 2A
03_monte_model_probe_subset	Assess how different probe subsets affect purity prediction	Figure 2C
04_monte_correlation_between_purity_metrics	Assess the correlation between predicted purity and other purity metrics	Figure 2B
05_benchmark_cancer_cor	Compare purity predictions from different methods	Figure 3A
06_monte_pancancer_probe_correction	Train the model and perform probe correction	Figure 5A
07_benchmark_methylation_correction	Compare probe-correction methods	Figure 5B
08_monte_pcawg_fine_tuning	Fine-tune the model on the PCAWG dataset and evaluate different training probe sizes	Figures 4A and 4B
09_application_cancer_correction	Fine-tune the pan-cancer model for a target cancer type	Cancer-specific models (released in the MONTE repository)
10_application_BRCA_marker	Identify BRCA markers after probe correction	Figure 6

Environment setup

Python environment

The Python environment for this repository is managed using uv. Before proceeding, make sure that Git and uv are installed and available from your terminal.

MONTE currently requires Python 3.13 or later.

After cloning this repository, enter the project directory and synchronize the environment:

cd MONTE-analysis
uv sync --locked

This command creates a virtual environment in .venv and installs the exact dependency versions recorded in uv.lock, including the MONTE package.

If you use Jupyter through an editor such as VS Code, select the Python interpreter located in the .venv directory.

Verify the installation

You can verify that MONTE was installed correctly by running:

uv run python -c "import monte; print('MONTE was imported successfully.')"

If this command completes without an error, the Python environment is ready.

For information about installing and using MONTE independently of this analysis repository, see the MONTE repository.

R environment

The R packages used by the R scripts are not managed by uv. Install the required packages in your R environment before running these scripts.

Running the analyses

Download the TCGA and PCAWG methylation datasets from Zenodo, and move the data folder to the current directory.

The analyses can then be run using the notebooks and scripts provided in this repository.

Citation

The MONTE preprint is available on bioRxiv.

If you use MONTE or the code in this repository in your research, please cite:

@article {Kim2026.01.22.701164,
    author = {Kim, Mirae and Lee, Wei-Hao and Yao, Vicky},
    title = {MONTE enables unified pan-cancer tumor purity estimation and methylation correction from bulk DNA methylation arrays},
    elocation-id = {2026.01.22.701164},
    year = {2026},
    doi = {10.64898/2026.01.22.701164},
    publisher = {Cold Spring Harbor Laboratory},
    URL = {https://www.biorxiv.org/content/early/2026/08/26/2026.01.22.701164},
    eprint = {https://www.biorxiv.org/content/early/2026/08/26/2026.01.22.701164.full.pdf},
    journal = {bioRxiv}
}

Contact

If you encounter a problem with this repository, please open a GitHub issue or contact Wei-Hao Lee at wl61@rice.edu.